coreSR.AI
Learn

EviTrace
Evidence retrieval engine

AI-guided question-framework extraction, Boolean strategy generation, open-database search, and citation graph expansion

EviTrace walks you from a research question to a screening-ready reference set in four steps. It searches open-access databases, checks seed-study recovery, and generates strategies you can run in licensed databases like MEDLINE and Embase.

92.72%frozen-output study recall395 / 426included studies recognized19Cochrane development/regression reviews

EviTrace search workspace

EviTrace searches open-access databases (PubMed, Europe PMC, Crossref, ClinicalTrials.gov, Semantic Scholar) and can optionally add OpenAlex citation-graph expansion with the configured service key or your own run-time API key. It does not replace searching licensed databases directly. For a comprehensive, publication-quality systematic review, you must also search MEDLINE via Ovid, Embase, CENTRAL, and other relevant databases through your institution. Use the AI-generated search strategy as a starting point for those searches.

How It Works

Four steps from research question to screening-ready references.

1DefineQuestionChoose frameworkEnter criteria2SeedPapersKnown studiesPMID lookup3AI SearchStrategyMeSH termsBoolean syntax4ExecuteSearchOpen-access DBsDeduplication5ScreeningBatchFramework retainedReady to screen

Step-by-Step Guide

Each step saves automatically so you can close the browser and return later.

Step 1

Define Your Review Question

Choose PICO (intervention), PECO (exposure), PICOTS (prognosis), PIRD (diagnostic), CoCoPop (prevalence), PICo (qualitative), or PCC (scoping), then enter the framework-specific criteria manually or extract them from a protocol.

Be specific and expansive. Include synonyms, alternate names, and subtypes for the framework’s key concepts. The more detail you provide, the better the AI search strategy.
Step 2

Add Seed Papers (Optional)

Add known relevant studies by PMID or title search. Seed papers help the AI understand exactly what kind of studies your review should capture, improving search strategy quality.

Even 2-3 seed papers significantly improve search strategy generation. Use landmark trials or highly relevant systematic reviews.
Step 3

Generate AI Search Strategy

The AI analyzes your framework-specific criteria and seed papers to generate a Boolean search strategy with MeSH terms and keywords. Strategies are generated in PubMed/MEDLINE syntax and general Boolean format.

Always review and refine the generated strategy. You can edit both the PubMed-specific and general versions before executing.
Step 4

Execute Search & Screen

Run the search across selected open-access databases, optionally add OpenAlex one-hop citation graph expansion, then import deduplicated results into a screening batch with the selected framework and criteria intact.

After running the preliminary search here, run the same strategy in MEDLINE, Embase, and other institutional databases for completeness.

Database Coverage

EviTrace searches these databases automatically. For a comprehensive review, you must also search licensed databases through your institution.

Open-Access Databases (Searched Automatically)

DatabaseCoverageNotes
PubMed36M+ biomedical citationsOverlaps with MEDLINE; uses E-utilities API
Europe PMC44M+ life-science citationsIncludes preprints from bioRxiv/medRxiv
OpenAlex graph250M+ works across all fieldsOptional one-hop citation graph add-on; uses a configured service key or your run-time key
ClinicalTrials.gov500K+ registered studiesTrial registrations, not published articles
Semantic Scholar200M+ papersAI-enhanced metadata; good for CS/bio overlap

Institutional Databases (Requires API Key)

DatabaseCoverageNotes
Scopus94M+ recordsRequires Elsevier API key
Web of Science90M+ recordsRequires Clarivate API key

Databases You Must Also Search Separately

These databases require institutional access and cannot be searched through EviTrace. Run your search strategy in each of these, then import the results as RIS or CSV files into the Reference Screener.

DatabaseWhy It Matters
MEDLINE via OvidGold-standard MeSH indexing; broadest controlled vocabulary search
EmbaseDrug/pharmacology focus; Emtree indexing captures drug names PubMed may miss
CENTRALCochrane’s curated register of controlled trials
CINAHLNursing and allied health literature
PsycINFOPsychology and behavioral sciences
Web of ScienceCitation tracking and multidisciplinary coverage
Grey literature sourcesDissertations, conference proceedings, reports

AI Search Strategy Generation

The AI generates structured Boolean search strategies following Cochrane and PRESS guidelines.

PubMed/MEDLINE Syntax

MeSH terms, field tags ([MeSH Terms], [tiab]), Boolean operators, and subheading qualifiers. Ready to paste into PubMed or MEDLINE via Ovid.

General Boolean Syntax

Database-agnostic Boolean query suitable for Embase, CINAHL, PsycINFO, Scopus, and other databases. Adapt controlled vocabulary terms to each database.

PRESS Compliance

Strategies follow the Peer Review of Electronic Search Strategies (PRESS) checklist principles: concept mapping, controlled vocabulary, free-text terms, Boolean logic, and filter application.

Best Practices for Search Strategies

  • Always review the AI-generated strategy before executing. Check that MeSH terms are current and that synonyms cover the key concepts in your selected framework adequately.
  • Use the generated strategy as a starting point. Adapt it for each database’s specific controlled vocabulary (MeSH for PubMed/MEDLINE, Emtree for Embase, CINAHL headings for CINAHL).
  • Run the strategy in at least two major databases (e.g., MEDLINE + Embase) in addition to the open-access sources searched here.
  • Document your search strategy, databases searched, and date of search for PRISMA compliance and reproducibility.
  • Consider having a medical librarian or information specialist peer-review your final strategy before executing the definitive search.

Recommended Workflow

How EviTrace fits into a complete systematic review search process.

1

Use EviTrace

Choose the review framework, define its criteria, generate your search strategy, run an open-access database search, and optionally add OpenAlex graph expansion.

2

Run in licensed databases

Copy the generated PubMed/MEDLINE strategy into MEDLINE via Ovid. Adapt the general strategy for Embase, CENTRAL, CINAHL, and other databases relevant to your topic.

3

Export and import

Export results from each licensed database as RIS files. Import them into the Reference Screener alongside the EviTrace results.

4

Deduplicate and screen

The Reference Screener automatically deduplicates all imported references. Your selected framework and field meanings carry forward from EviTrace into screening.

Frequently Asked Questions

Common questions about EviTrace and database coverage.

Does this replace searching MEDLINE, Embase, or CENTRAL directly?
No. EviTrace searches open-access APIs such as PubMed, Europe PMC, Crossref, ClinicalTrials.gov, Semantic Scholar, and optional OpenAlex citation-graph expansion when enabled. CoreSR can use a configured service key when available; a user-supplied OpenAlex key can override it for a single run and is not saved. For a comprehensive systematic review, you must also search subscription databases like MEDLINE via Ovid, Embase, CENTRAL, and others through your institution.
Why does CoreSR not search MEDLINE/Ovid or Embase directly?
MEDLINE via Ovid, Embase, CENTRAL, and other licensed databases require institutional subscriptions and use proprietary search interfaces. Their licensing terms do not permit third-party API access for automated retrieval. We provide the search strategy in proper syntax so you can run it yourself in those databases and import the results back.
How is this different from PubMed search on the PubMed website?
EviTrace queries the PubMed E-utilities API programmatically, which returns the same records as the PubMed website. The advantage is that results are automatically deduplicated against other sources, optionally expanded through citation-graph neighbors, formatted for screening, and imported directly into a screening batch with your selected question framework and criteria pre-configured.
Can I use the AI-generated search strategy in other databases?
Yes. The AI generates strategies in both PubMed/MEDLINE syntax and a general Boolean format. Copy the PubMed syntax into PubMed or MEDLINE via Ovid, and adapt the general syntax for Embase, CINAHL, PsycINFO, or any other database. Always review and adapt the strategy to the specific database’s controlled vocabulary (e.g., Emtree for Embase).
What does the AI question-framework extraction do?
After you choose a review type, the AI extracts the matching fields from an uploaded protocol: PICO for interventions, PECO for exposures, PICOTS for prognosis, PIRD for diagnostic accuracy, CoCoPop for prevalence, PICo for qualitative evidence, or PCC for scoping. PICOTS also captures prediction timing and setting. You can edit every field before proceeding.
How does deduplication work?
Results from all selected databases are merged and deduplicated using DOI matching first, then PMID matching, and finally fuzzy title matching for records without standard identifiers. This typically removes 15-30% of duplicate records.
Can I add Scopus or Web of Science results?
Yes. If you have Scopus (Elsevier) or Web of Science (Clarivate) API keys, you can add them in Account Settings. EviTrace will then include these databases alongside the free sources. You can also export your results from Scopus/WoS manually and upload them as RIS files in the Reference Screener.
Is there a limit to how many references can be returned?
Yes, but the limits differ by retrieval lane. Seed-review bibliographies are harvested aggressively because those references are expected to enter the candidate set. Forward citation expansion, auto-selected graph seeds, and broad open-source searches remain capped to control runtime, API use, and screening volume. For very broad questions, run complementary searches in licensed databases and import RIS files.
What is a seed paper and why should I add them?
Seed papers are studies you already know should be included in your review (e.g., from a prior scoping search or a seminal trial). They help the AI generate a more targeted search strategy by extracting relevant MeSH terms, keywords, and Boolean structure from papers that exemplify your research question.
My review needs to follow PRISMA guidelines. Does EviTrace help?
EviTrace records which databases were searched, the exact query used, and the number of results per source. This data forms the basis of your PRISMA flow diagram. For full PRISMA compliance, you should also document your searches in licensed databases (MEDLINE, Embase, etc.) which you run separately.

Ready to build your review?

Choose the framework that fits your review and let the AI generate your search strategy.

Start EviTrace

References

1. McGowan J, Sampson M, Salzwedel DM, et al. PRESS Peer Review of Electronic Search Strategies: 2015 Guideline Statement. J Clin Epidemiol. 2016;75:40-46. doi:10.1016/j.jclinepi.2016.01.021

2. Rethlefsen ML, Kirtley S, Waffenschmidt S, et al. PRISMA-S: an extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews. Syst Rev. 2021;10:39. doi:10.1186/s13643-020-01542-z

3. Higgins JPT, Thomas J, Chandler J, et al. Cochrane Handbook for Systematic Reviews of Interventions. 2nd ed. Wiley; 2019.

CoreSR · Evidence Synthesis Platform